Lich, calibration does nothing and output is offset

OWL Lich
,
#1

Hi, I’m new to Lich and looking forward to writing some code for it.
Thinking that a nice patch to have available would be one to test all ins and outs, and allow user to set calibration by ear.

Anyway, first need to actually get it working.

I’m able to upload patches to the Lich, and everything seems to be working, apart from calibration.

I can run the calibration, but using the check High voltage measured stays at 4.99, and low at -4.88.
Doesn’t matter how many times i calibrate, or even if I type in silly values for voltage, it stays the same.
Web page shows ‘test results’ of 5.01 and -5.01.

Using the calibrate ( and lead from out to in) patch I measured 0.02V with neither button pressed.

Just measuring the output, without the connection to input, I get
High button 5.05V
Low Button -5.02V
no button press 0V

Any ideas?

I used the “Test Calibrate” patch also as a voltmeter, just patching a voltage to the input and hitting the web button to read it. This reads the same as my voltmeter. ( generated 4V and -4V with a Doepfer precision adder, voltmeter +3.98, -3.98. Lich reads +3.97 and -3.98. I don’t vouch for either the accuracy of the Doepfer module, to which I calibrate my system, or the voltmeter which is old and cheap…but it makes me think that the input is ok.) Input reads 0V on the webpage calib check with no connection.

#2

Hi Andy,
Is your multimeter yellow or orange?..

Have a look at this video with calibration details if you haven’t seen it. And maybe this thread including the comment about adding resistance for calibration.

If neither helps, we can try summoning @mars here.

#3

Many thanks.
I now have a module calibrated to the poor standard of my meter, which is grey actually.
Main thing is :-
The “Test High” and “Test Low” buttons on the webpage are not the same as buttons on the Lich
(from the alt calib vid which you suggested)
(not sure about the impedance thing , but did it anyway)

Other ‘learning curve’ from today is that you have to use chrome, and not Firefox with midi enabling add-ons.

At least it works, and accurate calibration can wait.

Thanks again :slight_smile:

1 Like
#4

So calibration is only really possible with an accurate meter, and number of figures shown on the meter is no guarantee that the meter is accurate.
Plus, a lot of people calibrate by ear to a single device, i use a Doepfer precision adder for reference voltages.
So choice is likely to be, get an expensive meter and re-calibrate your whole system to that.

OR
Use you crappy meter (which is perfectly good enough for most stuff) to measure the reference voltages on your system.
Work out the error your meter is showing when measuring your reference voltages.
Calculate a fiddle factor, so when you do Lich calibration you fool the webpage by modifying the measured voltage. e.g. I measured noninal 4V on my (in tune) system with my cheap meter and got 3.98V
(I don’t have 5V reference, that would have been better)
Then when reading the high calibration (+5V) on the Lich I added 0.02 volts to the figure before entering it.
Similar for the low calib.
Now the Lich is pretty closely tuned to my system.

Would be nice if Lich could be calibrated by ear

#5

I’ve made something similar to what you describe in Magus firmware a while ago. The idea was to calibrate based on input voltage. It received 1V / 3V and compute scalar/offset values based on that, DMM wasn’t required. But it has been removed since than and doesn’t look like we’re getting a replacement.

#6

hmm yes or with something like a guitar tuner.

I’m sure we could do this, what would be the best way?
If we make a sine oscillator fixed at say 440Hz plus/minus whatever CV voltage it is seeing, then you could send it a known voltage and adjust the tuning to suit.
Would need at least two points: 0V to calculate offset, and then a high enough voltage to accurately set the scalar.
All we need is a simple oscillator patch that continuously requests the tuning coefficients from the firmware (all existing patches do this once when they are loaded), and a web MIDI page with offset and scalar controls on it.

#7

Which works well when you’ve got accurate reference voltages, right?

We could bring it back as an alternative tuning method in the form of another web MIDI calibration page, where you could choose your two reference voltages. I don’t think it belongs in the firmware though.

#8

Accurate voltage source would be necessary only if you need to use OWL itself as an accurate voltage source. It’s a “what you tune to is what you get” approach, meaning that if you have inaccurate pitch CV source, your OWL starts tracking the same way. So in theory you could use a sequencer with any amount of inaccuracy and it gets masked by matching scalar/offset values for input (just what Andy tried to achieve with current tuning code). Of course that error propagates to output too if you want to generate V/Oct signal on output.

My idea was that it should be possible to tune Magus or anything with display without any web pages, that’s why it was done in FW. If there’s no on-device tuning, I don’t particular care about an alternative tuning method.

Second goal for that Magus tuning was to be able to do it without DMM, that’s why it connected output to calibrated input for calibrating outputs. It could be modified to use DMM for outputs to avoid propagation of input calibration errors. For on-device calibration that would require entering 2 voltage readings.

#9

Isn’t it always going to be a “what you tune is what you get approach” ?
If you use a multi-meter, then it depends on the multi-meter.
A “three digit voltimeter” doesn’t even guarantee that those 3 digits are accurate. (mine isn’t).

I use the doepfer A-185-2 Precision adder to create reference voltages, I trust that more than the meter,
(Just tried it with my self calibrating Moog synth, and it creates in tune octaves, so I’m kind of stuck with it anyway).

Wish list (potentially todo list)

For output would be nice to have knobs for
Select octave
Scale
Offset (not essential)
…saveable from a patch that doesn’t need a webpage.
Ideally would want it for both left and right.
so that would be usable with “your favourite oscillator”, or with a meter.